: Vocabulario de PLN: reinforcement learning

NLP methods and tools > artificial intelligence > machine learning > machine learning approach > reinforcement learning

reinforcement learning

A subset of machine learning that allows an AI-driven system (sometimes referred to as an agent) to learn through trial and error using feedback from its actions. This feedback is either negative or positive, signalled as punishment or reward with, of course, the aim of maximising the reward function. (University of York website)

Before reinforcement learning (RL) supervised learning (SL) is applied to mimic dialogues provided by a rule-based system. (Wang, Zhang, Li, Zong & Li, 2019)
During the action sampling step in RL we reduce the search space of actions based on the constitution of the previous word contexts as well as our n-gram model. (Guo, Chang, Yu & Bai, 2018)
Some recent studies have incorporated RL to align LMs with human preference and to prompt LM for problem-solving (see Table 1 for details). (Zhou, Du & Li, 2024)
This model is developed using the previous model by adding mixed training of both machine learning and reinforcement learning. (Firdaus, Ekbal & Bhattacharyya, 2020)

http://data.loterre.fr/ark:/67375/8LP-B7HRCSC1-T

RDF/XML TURTLE JSON-LD última modificación 29/5/24

Loterre